The Journal

Synaptic Physiology Cortex Judith

A. Hirsch and Charles

Laboratory

of Neurobiology,

of Horizontal

Connections

of Neuroscience,

June

1991,

1 I(6):

1800-l

809

in the Cat’s Visual

D. Gilbert

The Rockefeller

University,

New York, New York 10021

Horizontal connections are a principal component of intrinsic cortical circuitry. They arise mainly from pyramidal cells and course parallel to the brain’s surface for distances as long as 8 mm, linking columns with shared orientation preference and allowing cells to integrate visual information from outside their receptive fields. We examined the synaptic physiology of the horizontal pathway in slices of the cat’s striate cortex and found that activating lateral fibers produced both excitation and inhibition. We recorded the postsynaptic responses of identified pyramidal cells in layer 2+3 of area 17 to electrical shocks applied at three sites: in the home column of the impaled neuron either in layer 2+3 or 4, or at a lateral distance of 0.9-3 mm in layer 2+3. Within the home column, suprathreshold stimuli produced compound EPSPs with action potentials, followed by fast, GABA,ergic IPSPs and a slower, GABA,ergic IPSP. For the distant stimulating site, the threshold response was an EPSP. Stronger shocks frequently evoked a disynaptic, GABA,ergic IPSP that truncated the EPSP and could dominate the postsynaptic response. At the resting potential, the horizontally evoked EPSP was too small to elicit spikes. With depolarization of the membrane, however, it grew several hundred-fold. This amplification was blocked by N-(2,8-dimethylphenylcarbamoylmethyl)triethylammonium bromide @X-314), but not by 2-amino+phosphonovalerate (APV), indicating that it was mediated by Na+ channels, rather than by NMDA receptors. We propose that the horizontal connections provide the means for stimuli outside the receptive field to modulate activity elicited within its confines. The voltage-dependent enhancement of the laterally evoked EPSP may explain why stimulating the surround by itself fails to drive cells but can facilitate their response to stimuli within the receptive field. The ability to initiate disynaptic inhibition from lateral sites shows that recruiting appropriate groups of horizontal fibers can also have a suppressive effect. Thus, the effect of horizontal input is state dependent, with the size and sign of the laterally evoked response changing according to the balance of converging inputs. Received Oct. 17, 1990; revised Jan. 24, 1991; accepted Jan. 30, 199 1. We are grateful to Torsten N. Wiesel for his support throughout the project and for his careful and critical comments on the manuscript. We also thank Kaare Christian for help with computers, Peter Pence for photography, and Shari Zagorski for technical assistance. These experiments were funded by NIH Grants EY06010 to J.A.H. and NS22789, National Science Foundation Grants BNS89 1895 1 and BNS835 1738, and an award from the Rita Allen Foundation to C.D.G. Correspondence should be addressed to Judith A. Hirsch, Laboratory of Neurobiology, Box 138, The Rockefeller University, 1230 York Avenue, New York, NY 10021. Copyright 0 1991 Society for Neuroscience 0270-6474/91/l 11800-10$03.00/O

The cerebral cortex transforms information coming from the thalamus and other cortical areasby meansof a stereotypedset of vertical and horizontal connections. In the primary visual cortex, cells with shared receptive fields are connected by interlaminar, vertical collaterals and are arrangedin radially oriented columns (Lorente de No, 1944; Hubel and Wiesel, 1963). Another, more recently discovered set of connectionsare longrange,clustered, horizontal projections that coursefor 6-8 mm within a single layer (Gilbert and Wiesel, 1979, 1983, 1989; Rockland and Lund, 1983;Martin and Whitteridge, 1984).These extensive collaterals link columnswith sharedorientation preferenceand enable recipient cells to integrate information over a larger visuotopic areathan the “classical” receptive field, that region in which a simple stimulus can evoke activity (Ts’o et al., 1986; Gilbert and Wiesel, 1989). The current study explores the synaptic physiology of the horizontal, or “lateral,” pathway, in order to help learn how information from outsidethe classical receptive field can modulate a neuron’sresponse. Stimulation of the surround has both inhibitory and excitatory influenceson responsesevoked from the central part of the receptive field (Allman et al., 1985;Nelsonand Frost, 1985; Orban et al., 1987;Wieseland Gilbert, 1989;Gilbert and Wiesel, 1990). These influences could be mediated by lateral inputs (Wiesel and Gilbert, 1989; Gilbert and Wiesel, 1990). Though the horizontal collateralsarisemainly from pyramidal cellsand usually contact other pyramids, as many as 20% of the postsynaptic targetsare smoothcells(I&v&day et al., 1986;McGuire et al., 1991). Thus, stimulating horizontal fibers could have disynaptic inhibitory effectsaswell asdirect excitatory actions. Cross-correlation analysis indicates that lateral inputs are primarily excitatory. This is not, however, a technique well suited to resolve inhibitory interactions (Ts’o et al., 1986). In the present study, we recorded intracellularly from brain slicesto examine the physiological properties of the synapses in the horizontal pathway. Intracellular electrodes provide a meansof measuringsynaptic efficacy and of characterizing excitatory and inhibitory input. Furthermore, surgicaland pharmacological isolation of various presynaptic pathways is more easily achieved in vitro than in situ. Our resultsprovide insights about the excitatory and inhibitory mechanismsunderlying the role that horizontal connectionsplay in visual cortical processing.

Materials

and Methods

Anesthesia and surgery.Twenty-nineadult cats, 2-4 kg, were first anesthetized with an intramuscular injection of ketamine (10 mg/kg), after which anesthesia was maintained with intravenous Pentothal (20 mg/ kg). The EKG and temperature were continually monitored; temperature was controlled with a heating pad.

Theanimalwasplacedin a stereotaxicdevice,andtheskulloverlying the occipitalcorticeswasremoved.The dura wasincisedalongthe

The Journal

anteroposterior axis and reflected to expose the lateral gyrus. A block of the gyrus was then cut with a scalpel and removed with a spatula. The excised tissue was placed in oxygenated artificial cerebral spinal fluid (ACSF) for several minutes until slices were made. Preparation and maintenance of slices. After gently removing the pia from the gyrus, the block was placed on a slab of agar, usually with the medial bank facing upwards, and transferred to a tissue slicer modeled after a guillotine (Katz, 1987). This apparatus carries a rack containing about 50 gold-plated tungsten wires, 20 pm in diameter, that are arranged in parallel 400 Frn apart. The rack is suspended by a springloaded carrier and moves quickly through the tissue when the spring is released, producing up to 50 coronal slices. The slices were placed in ACSF, separated with a fine brush, and moved one by one to a closed interface chamber held at 25-30°C where they were kept for 2-28 hr. Their lower surfaces rested on nylon nets placed over a bath containing ACSF, and their upper surfaces were exposed to an atmosphere of 95% 0, and 5% CO, saturated with H,O. For recording, a slice was submerged in a glass-bottomed chamber with a volume of 0.4 ml and was held stable by a weighted nylon net. Control and test solutions were perfused through the chamber at a rate of ~0.5 ml/min and heated to 34-35°C. The chamber was illuminated from below and viewed from above through a dissecting microscope. In some experiments the superficial layers of the slices were isolated by undercutting with fragments of razor blade before the net was put in place. Recording and stimulation. Intracellular electrodes were pulled from borosilicate, capillary glass tubing with a 1.2-mm outer diameter and 0.6-mm inner diameter. Electrodes had resistances of 40-100 MB when filled with the electrolyte. Transmembrane voltage was monitored, and intracellular current was delivered through a bridge circuit and displayed on an oscilloscope and computer terminal. Data were digitized on line and stored on videotape and on disk for future analysis. Electrical stimuli consisted of shocks lastina from 50 to 150 Irsec, of between 1 and 100 V, that were delivered at Oyl Hz through two sharpened, insulated tungsten wires with a tip separation of ~50 pm. SolutionxThe standard ACSF contained. in mM. NaCl. 125: KCl. 5: NaHCO,, 126; CaCl,, 2.4; MgCl,, 1.3; and dextrose; 10. Itwas saturated with 95% 0,/5% CO, and adjusted to pH 7.4. All externally applied drugs were dissolved in the ACSF and delivered in the bath. The 2-hydroxy-saclofen and the 5-amino-phosphonovalerate (APV) were from Tocris Ltd.; the bicuculline methiodide (BMI) was from Research Biochemical Inc. The recording electrodes were filled with 3-4 M potassium acetate (KAc; pH, 7.2). At times, the electrode tips were backfilled with 2% biocytin (Molecular Probes) in KAc before the shanks were filled with the electrolyte. The biocytin labeled the cells during recording, and they could then be identified after histological processing. For some experiments, electrode tips were backfilled with 100 mM N-(2,6-dimethylphenylcarbamoylmethyl)triethylammonium bromide (QX-3 14; gift of Astra Pharmaceuticals) dissolved in 3 M KAc. Histology. The tissue containing biocytin-labeled cells was processed using a modification of the technique described in Horikawa and Armstrong (1988). Briefly, slices were fixed overnight in 4% paraformaldehvde. cut into 70-wrn sections with a freezing microtome, and rinsed twice with PBS. The sections were then incubated for 45 min in PBS containine 0.25% Triton and 2% BSA. After rinsing in PBS with 2% BSA, they were incubated in avidin-HRP (Vector Laboratories; diluted 1: 100 in PBS and 2% BSA) for 2 hr at room temperature or overnight at 4°C. The tissue was then washed in PBS (3 x 15 min) and incubated for 1 hr in 0.05% diaminobenzidine (DAB) in PBS. Hydrogen peroxide (0.3%) was then added at a ratio of 1: 100 to the DAB solution. About 15 min later, the sections were rinsed three times in PBS and were mounted, dehydrated, and coverslipped.

Results The resultspresentedherewereobtained from impalementsthat lasted from 20 min to 6 hr; our sampleincludes intracellular recordingsfrom 61 neurons. The average input resistancewas 41 ? 16 MB (mean -t SD), and the average resting potential was 77 -t 7 mV. Of the 18 cells labeled with dye, all were pyramids; no smooth stellate cells were filled. All the stained cells had action potentials whose width at half the maximum amplitude was 2 1 msec. Unlabeled neurons were identified

of Neuroscience,

June

1991,

1 l(6)

1801

physiologically either aspyramids if their action potentials lasted I 1 msecat half-height or assmoothcellsif their spikeswere 10.6 msecat half-height, accordingto the criterion of Schwartzkroin and Mathers (1978) and others (McCormick et al., 1985: Naegeleand Katz, 1990). The top right panel in Figure 1 showsa schematicdiagram of a typical slice. The recording electrode was placed in layer 2+3 while a stimulating electrode was placed at one of three sites:in near vertical alignment (within 50 pm) with the site of impalement, either in the same layer or in layer 4, or at a horizontal displacementof 800 pm to 3 mm in layer 2+3. The records were obtained from a single physiologically identified pyramidal cell. Stimulation from a distance(Fig. 1A) consisted only of an EPSP. In contrast, the synaptic potentials elicited in vertical register with the recording electrode (Fig. lB,C’) comprised a compound EPSP with action potentials followed by early, fast and then late, slow inhibitory potentials. While the vertical responsesevoked from layer 2+3 (Fig. 1B) were less robust than thosefrom layer 4 (Fig. 1C), they werequalitatively similar. The recordings in Figure 2 were made from physiologically identified pyramidal cells in three preparations where the superficial layers had beenundercut to remove ascendinginputs. Figure 2, A and B, illustrates both the excitatory and inhibitory extremesof the rangeof responses to stimuli at a distance(“far”; seeFig. 1, site A, upper right). The traces in Figure 2A, top to bottom, show responsesthat remained largely excitatory as stimulus strength was raisedfrom threshold to the maximum. The EPSPrecordedat threshold wassmall, lessthan 2 mV, and grew gradually over the spanof shock voltages; three examples are shown in Figure 2A. Even at the highest stimulus voltage, the EPSPswere too small to elevate the membrane from the resting potential to the threshold for firing (not shown). For 17 out of 29 cells in the undercut preparations, stimulation at a distanceevoked inhibition aswell asexcitation (Fig. 2B); the rest showedonly excitation. The sameslicecould contain neurons with both types of response.Though for the cell in Figure 2B the threshold responsewas mainly excitatory, increasingthe stimulus voltage evoked an early, fast IPSP that carved into the EPSP and eventually dominated the postsynaptic responseat high shock strengths. The averagedvalue of the latency of all the remotely evoked EPSPsdivided by the distance from the stimulus site was 3.5 f 1.2 msec/mm.

Because this value includes

both the synaptic

delay and spikeinitiation time (totaling x0.6-0.8 msec),it probably underestimatesthe conduction velocity of the horizontal fibers.Of 10cellsfor which the beginningsof both the excitatory and inhibitory potential were clearly distinguishedby a steep slope, the lag between the EPSP and IPSP, measuredat the maximum stimulus strength, was 2.6 ? 0.9 msec. This delay suggeststhat inhibition was, at the minimum, disynaptic. “Far” stimuli failed to evoke late IPSPs, unless,as will be shownlater, fast inhibition wasblocked with bicuculline. When serialEPSPswerepresentduring perfusionof the standardACSF, the early EPSPwasable to follow stimulation at high frequency, 1.7 Hz, while the later oneswere not, indicating that the first EPSPwasmonosynaptic while the subsequentoneswere di- or multisynaptic. When the stimulating electrode was placed within 50 pm of the impalement (“near”; seeFig. 1, site B, upper right), weak shockselicited an EPSP (Fig. 2C). At slightly higher stimulus voltages, the EPSP was cut short by a fast IPSP, as in Figure

1802

Hirsch

and

Gilbert

* Synaptic

Physiology

of Horizontal

Connections

B Figure 1. Synaptic potentials recorded from a single cell in layer 2+3 following stimulation of various sites. A schematic diagram of the preparation is shown in the upperright.Traces show EPSP evoked by a shock to layer 2+3, ~900 pm away from the stimulating electrode (site A) and EPSPs and fast and slow IPSPs following a shock to layer 2+3, 50 pm away (siteB) and from layer 4, 1 mm below the recording electrode (site C). Two traces are shown for each stimulus condition (A-C’), and action potentials in the lowertrace3are cropped. The membrane voltage was depolarized to -62 mV from its resting potential of -80 mV during the recording in order to visualize IPSPs. Arrowheads mark the stimulus artifact.

2A. A further increase in the shock strength evoked an action potential and a later, longer IPSP, as well. Overall, the “near” responseswere more robust than the “far” ones. The EPSPs werelargeenoughto drive action potentialsfrom rest(not shown), and both fast and slow IPSPs could always be evoked. Analysis of the laterally evoked EPSP During the course of our study, we noticed that the evoked EPSPs were much smaller when the membrane

was at rest than

when it wasdepolarized. This voltage dependence,which is the oppositeof the conventional one, where synaptic potentials decreasein size as their reversal potential is neared, has been observedat other central synapses.Sometimesthis phenomenon has been attributed to inward rectification due to noninactivating Na+ channels(Staftstrom et al., 1985, Sutor and Hablitz, 1989a,b)and, at other times, to the NMDA subtype ofglutamate receptor (Jonesand Baughman, 1988;Artola and Singer, 1990). We testedboth of thesepossibilitiesin our preparations. The voltage-dependentenhancementof the laterally evoked EPSPis pictured for a singlecell in the top panel of Figure 3A. When the membrane was at rest (Fig. 3A, top panel, bottom trace), the EPSPwas smaller and briefer than when the membrane wasdepolarized from its resting potential to several mil-

livolts below the threshold for spontaneousfiring (Fig. 3A, top panel, bottom trace). The current-voltage relationship for the neuron is plotted in the bottom of Figure 3A along with an inset showingan action potential fired in responseto injected current. Note that the plotted curve showsinward rectification, increasing in slope as the membrane potential was depolarized from rest. When a different cell in the samepreparation was impaled with an electrode containing QX-314 (Fig. 3B), a compound that blocks Na+ currents (Connors and Prince, 1982; Staftstrom et al., 1985),the EPSPassumeda conventional relationshipwith voltage, decreasingin size with depolarization (Fig. 3B, top panel, top trace). Thus, it seemedthat when the membrane potential neared the threshold for the activation of Na+ channels, as in Figure 3A, the resulting currents amplified the horizontally evoked EPSP(n = 3). The bottom panel of Figure 3B showsthe current-voltage relationship and shapeof the action potential of the cell injected with QX-3 14. The inward rectification and fast spike seenin the control situation were absent, confirming the effective block of Na+ channels.The slow action potential that remained was probably mediated by calcium (Connors and Prince, 1982; Staftstrom et al., 1985). Both cells in Figure 3 were labeled pyramidal cells.

The Journal of Neuroscience,

June 1991, 7 I(6) 1603

medium

Near

. medium 7 J

-

20 ms

We next examined the participation of NMDA receptors in the laterally evoked response. These are of particular interest because their activation correlates with long-term modifications in synaptic strength (Kauer et al., 1988). The traces in Figure 4 were recorded from a labeled pyramidal cell before (Fig. 4a), during (Fig. 4b), and after (Fig. 4c) the inclusion of 30 I.LMD-APV, an antagonist of NMDA receptors (Watkins and Evans, 1981) in the bath. The APV caused a slight decrease in the size of the EPSP, but did little to reduce its enhancement by depolarization (n = 4). The size of the voltage-enhanced EPSP shown after removal of the drug is smaller than it was at first, probably because the input resistance of the cell had decreased during recording. In sum, at potentials below the threshold for firing, it appears that Na+ currents are sufficient to explain the increase in the EPSPs’ amplitude with membrane voltage, with little or no contribution from currents gated by NMDA receptors. Analysis of laterally evoked inhibition There is diversity among receptors mediating IPSPs as well as EPSPs. In cortex, GABA, receptors mediate fast, chloride-de-

Figure 2. Synaptic responses measured from the undercut layer 2+ 3 when the stimulating electrode was 1.2 mm (A, B) or 50 pm (C) away from the micropipette. Each trace is the average of three trials. A, Graded EPSPs recorded from a cell whose response to horizontal inputs was primarily excitatory. B, Synaptic response of a cell whose response to horizontal input also included a araded fast IPSP at stimulus voltages above threshold. C, EPSPs and higher-threshold fast and slow IPSPs evoked from a site near the impaled cell; in the bottom truce the response included an action potential, which is cropped. The recordings in each panel were made from three senarate cells. The membrane potential was depolarized from -76 to -58 mV for A, from -77 to -65 mV for B, and from -82 to -60 mV for C. The voltage ranges of the shocks used to stimulate the cells in this figure were as follows: low, 5-l 5 V, medium, 25-50 V, high, 15-100 V. Arrowheads mark the stimulus artifact.

pendent IPSPs, and GABA, receptors transduce longer-lasting IPSPs that rely on increased conductance to K+ (Newberry and Nicoll, 1985; Connors et al., 1988; Dutar and Nicoll, 1988a,b). We assayed for the contribution of both subtypes of receptor to the laterally evoked IPSPs by applying selective antagonists to the slices. The recordings displayed in Figure 5 were obtained from a physiologically typed pyramid. Each pair of traces shows two individual trials under the conditions labeled at the upper right. In the control medium, stimulation of lateral inputs evoked the small EPSP followed by a weak, fast IPSP (Fig. 5~). Adding a selective antagonist of GABA, receptors, BMI (25 FM), to the bath removed the fast IPSP from the cell and revealed a large burst followed by a slow hyperpolarization (Fig. 5b; n = 4). An antagonist of GABA, receptors (Kerr et al., 1988), 2-hydroxysaclofen(0.25-0.5 mM), suppressed the afterhyperpolarization, indicating that it was produced synaptically and that it was a slow IPSP (Fig. 5~; n = 2). The residual afterhyperpolarization that remained in 2-hydroxy-saclofen was likely a product of intrinsic, nonsynaptic currents. As shown in the inset (Fig. SC), firing produced by injection of depolarizing current was followed

1804

Hirsch

and Gilbert

* Synaptic

Physiology

of Horizontal

Connections

QX-314

L

5mV 20m.s

Figure 3. QX-3 14abolished the voltage-dependent enhancement of the lateral EPSP.The upper trace in each of the top panels is the average of three trials recorded when the membrane was depolarized by 15 mV, and the lower trace is averaged from trials at rest. For the control cell, depolarization increased the size of the EPSPthreefold

(A), but whenQX-3 14wasinjectedin another neuron, the size of the EPSP

was reduced by depolarization. Arrowheads mark the stimulus artifact. The bottom panels showthe action potentials and current-voltage relationships of the two neurons. Bothcellsidentified as pyramids in layer 2. The resting potentials of the cells on the right and left were -75 and -79 mV, respectively.

l-20 AV(mV)

by a similar hyperpolarization. The effects of both drugs were reversible (Fig. We). This experiment, as well as the results shown in Figure 2, suggeststhat while both types of receptors are presentin the undercut layers, fast GABA,ergic inhibition was primarily evoked by lateral inputs. The largebursts produced when fast inhibition was blocked contained polysynaptic EPSPs,as illustrated in Figure 6, obtained from a physiologically classifiedpyramidal cell. Thus, fast inhibition

strongly

limits

the effects of horizontal

inputs.

The threshold responsein the normal salinewasa small EPSP (Fig. 6~). After superfusionwith BMI, the strength of the shock required to evoke a synaptic potential dropped from 70 V (Fig. 6~) to 12 V (Fig. 6b), and the threshold responseconsistedof a train of EPSPs occurring at long latency. As stimulus voltage wasraised (Fig. 6c,d), the latency of the burst shorteneduntil, at 7OV, the rising phaseof the original, monosynaptic EPSPwas coincident with that of the burst. The shorteningof latency with stimulus

intensity

implies

that the excitatory

intemeurons

are

excited more rapidly aslarger numbersof convergent inputs are activated. Possiblythe appearanceof the long slow IPSP, asseen in the last figure, occurred becauseGABAergic intemeurons were alsomore strongly excited. The traces shownin this figure were recorded at rest to emphasizethe EPSPs. Consequently,

AV(mV)

the IPSPs were not visible becausethe membranevoltage was below or near their reversal potentials. The tracesin all the previous figureswere obtained from cells that were identified as pyramidal on either physiological or anatomical grounds.In rare instances,we recorded from cellsthat we physiologically classifiedas intemeurons; the responsesof these cells to lateral input were very different from those typically exhibited by pyramids. Recordingsfrom a presumedintemeuron are shown in the top rows of Figure 7, and responses obtained from a stained pyramidal cell are shown below; each pair of traces showstwo trials of the same stimulus. As the shock voltage increased, the interneuron fired progressively harder, showingthe pattern of increasingactivity that onewould expect to generatethe intensifying inhibition characteristically shown by pyramidal cells as stimulus strength is raised.

Discussion Our resultsshowthat horizontal connectionscan inhibit aswell asexcite their postsynaptic targets. In the following paragraphs, we discusstheselaterally evoked responsesin terms of what is known about their underlying anatomy and how they compare with synaptic potentials recorded in various cortical areas.Also,

The Journal

of Neuroscience,

June

1991,

f7(6)

1605

we consider our findings in the context of those from experiments that have explored horizontal connections with visual stimuli. in vivo. Source of the synaptic potentials The responses evoked from sites in vertical register with the impaled cell probably arose from the home column of the impaled neuron, while the distant sites were on the order of one to three hypercolumns away. A hypercolumn is a segment of cortex containing a full complement of orientation or ocular dominance columns and in the cat is approximately 1 mm wide (Hubel and Wiesel, 1963; Lowe1 et al., 1987, 1988). The terminals of the horizontal collaterals are distributed in clusters and connect columns of similar orientation preference (Gilbert and Wiesel, 1979, 1983, 1989; Martin and Whitteridge, 1984; Ts’o et al., 1986). As a consequence, one might expect that EPSPs produced by lateral inputs would arise only from spatially discrete loci arrayed at a periodicity roughly equal to the width of a hypercolumn. Our stimulus was a shock, however, so it could activate fibers of passage as well as the somata underneath the electrodes, making it difficult to identify the columns where the inputs originated. We are currently designing experiments to study the specificity of the horizontal inputs in the slice preparation. The most likely source of the synaptic potentials we recorded in the undercut preparations was input intrinsic to layers 2+3 of area 17. Callosal terminals are largely confined to the area 17/l 8 border (Hubel and Wiesel, 1967; Shatz, 1977; Innocenti, 1980), and most of the other inputs to the upper layers, such as thalamic afferents (LeVay and Gilbert, 1976; Humphrey et al., 1985) or the diffuse projections from the brainstem (Morrison et al., 1982) would have been transected. Descending projections from extrinsic cortical regions, areas 18 and 19, for example, may course through the superficial laminae (Creutzfeldt et al., 1977; Tigges et al., 198 1; Rockland and Virga, 1989), but are sparse relative to the fibers originating in area 17. Intracellular injections show that the majority of pyramidal cells, themselves accounting for 80% of the cellular population in layer 2+3, extend long-range collaterals (Gilbert and Wiesel, 1979, 1983, 1985; Martin and Whitteridge, 1984; Callaway and Katz, 1990). Excitation The laterally evoked EPSP was too weak to trigger action potentials, unless the membrane was depolarized from rest by the injection of current through the microelectrode. In view of previous studies, the small size of the lateral EPSP is not surprising, and many synchronously active fibers are probably required to exert even a small effect. The axonal arbors of the long-range collaterals are not nearly as dense as the intracolumnar terminal fields (Gilbert and Wiesel, 1979, 1983) and make few contacts on each postsynaptic cell (Kisvlrday et al., 1986, Gabbott et al., 1987; McGuire, 199 1). Comparison of cross-correlation studies of inter- versus intracolumnar connections indicates that the lateral ones are about an order of magnitude weaker (Toyama et al., 1981a,b; Ts’o et al., 1986; Toyama, 1988). Taken together, our own evidence and that of the earlier anatomical and physiological studies suggest that the role of horizontal connections is likely to modulate rather than generate activity within a given column.

I 5mV 20ms Figure 4. APV had little effect on the voltage-dependent enhancement of the lateral EPSP. Each trace is the average of three trials; the top truces were collected when the membrane was depolarized by 18 mV, and the bottom truces were taken at rest (- 82 mV). The application of 30 PM D-APV reversibly reduced the size of the EPSP but did not prevent its enlargement by depolarization. This cell was identified as a pyramid in layer 2. Arrowheads mark the stimulus artifact.

Voltage dependenceof the lateral EPSP An important feature of the EPSPproduced by horizontal fibers was its state dependence:it increasedin size as the membrane potential was raised from rest to near the threshold for spontaneousfiring. Voltage-dependentEPSPshave been reported in earlier cortical studieswhere synaptic responseswereapparently evoked from the home column, that is, from sitesnear or vertically aligned with the recording micropipette (Staftstrom et al., 1985; Jones and Baughman, 1988; Thomson et al., 1988; Sutor and Hablitz, 1989a,b; Artola and Singer, 1990). In most of thesereports, the early EPSPswere enhancedby depolarization, whereasthe later onesshowedconventional voltage relationships (Sutor and Hablitz, 1989a,b). From experiments usingvoltage clamp, injection of QX-3 14, or application of APV, Sutor and Hablitz (1989a,b) found that, in layer 2+3 of adult rat cortex, Na+ currents are the principal sourceof the augmentation of the early, intracolumnar EPSP. In contrast, others conclude that application of APV decreases the voltage dependenceof that EPSP (Jones and Baughman, 1988; Artola and Singer, 1990). There is agreementwith our present work, however, that at rest the NMDA-receptor-mediated component of the cortical EPSP is small (Jones and Baughman, 1988; Shirokawa et al., 1989; Sutor and Hablitz, 1989a,b;Artola and Singer, 1990). The voltage dependenceof the horizontally evoked EPSPis

1606

Hirsch

and Gilbert

* Synaptic

Physiology

of Horizontal

Connections

control

c JM/J ...... L

b

20mV 20ms

BMI OH-saclofen

BMI

BMI

40ms Figure 5. IPSPs mediated by both GABA, and GABA, receptors were generated by elements intrinsic to the undercut cortex. The fast IPSP (a) was removed by the application of 25 PM BMI, yielding an excitatory burst followed by a slow inhibitory potential (b) that was blocked by 0.5 mM 2-hydroxy-saclofen (c). The inset in c shows that an afterhyperpolarization such as that seen when the slow IPSP was blocked could be produced by direct injection of depolarizing current. The effects of both drugs were reversible (d,e). Two traces are shown for each stimulus condition, and action potentials in the lower truces are cropped. The broken lines indicate the baseline in the top traces of b-d. The traces were made when the membrane was depolarized from - 79 to -6 1 mV. Arrowheads mark the stimulus artifact.

a

Figure 6. The excitatory burst generated in bicuculline was produced by multisynaptic EPSPs. When fast inhibition was blocked, the threshold stimulus dropped from nearly 70 V (a) to 12 V (b), and the threshold response changed from a monosynaptic EPSP at short (3.6 msec) latency to large, suprathreshold potential at long (35 msec) delay. Increasing the shock strength shortened the latency of the burst (c) until it overlapped the monosynaptic EPSP at the original threshold voltage (d, trace in a shown here as a broken line). Each trace is the average of three trials recorded while the membrane was at rest, -83 mV. Arrowheads mark the stimulus artifact.

C

20 vq i

L

10 mV 20 ms

control

b

bicuculline

The Journal

of Neuroscience,

June

1991,

7 l(6)

1807

interneuron

Figure 7. Comparedresponse of a 10m 24ma pyramid

air”----

-‘-

low

high

not obviously different from the monosynaptic, intracolumnar ones, though it is unclear if the underlying mechanisms are the same. The meager size of the lateral EPSP, however, could make it particularly dependent on intrinsic depolarizing conductances or on additional synaptic excitation to bring the postsynaptic cell to fire. Because the amplification of the lateral EPSP does not appear to require NMDA receptors, one need not equate this behavior with the long-term changes associated with the NMDA receptor, at least within the voltage range tested here (cf. Hestrin et al., 1990). The voltage dependence may simply provide a means whereby the effect of the horizontal inputs would be subject to the context of levels of convergent activity. EPSPs in the presence of bicuculline

In our preparations, compound EPSPs evoked by horizontal stimuli were infrequent, but they always appeared when fast inhibition wasblocked by bicuculline. Field potentials recorded from layer 2+3 in cat and rat have shown that activity can propagatehorizontally for millimeters when inhibition is pharmacologicallyreduced(Chervin et al., 1988).Our resultssuggest that this propagationcan be aidedby elementswithin layer 2 + 3. Inhibition The marked role of inhibition in synaptic interactions within the superficial layers is perhaps unexpected given the statistic that no more than 20% of the terminals of layer 3 pyramids contact inhibitory cells(I&v&day et al., 1986; McGuire et al., 199l), yet many factors can influence the strength of connections, such asthe distribution of presynaptic contacts, the electrical characteristicsof the postsynaptic cell, and the arrangement of its boutons on the next neuronsin the synaptic net. In fact, smooth stellate cells are more densely innervated and receive proportionately fewer symmetric, presumably inhibitory, contacts than do pyramidal cells (McGuire et al., 1991). Furthermore, in cortex, GABAergic cellsfire more rapidly per unit

presumed inhibitory intemeuronanda pyramidalcellto lateralinput. The top truces madefroma presumed intemeuron showthat increasingthe stimulus strengthevokesprogressively stronger excitation.The reversesituationistrue for thepyramidalcell; strongerstimuli shift the balanceof the synapticinput towards inhibition. Tracesare displayedasin Figure 1; the membrane potentialwasdepolarizedfrom -63 to -60 mV for the top traces and from -85 to -62 mV for the bottom truces. Arrowheads markstimulusartifact.

increasein applied current than do pyramidal cells, a behavior that apparently reflects an actual difference in excitability between the two types of neuron in their balancesof K+, Na+, and Ca*+channels(Schwartzkroin and Mathers, 1978; McCormick et al., 1985; Giffin et al., 1988; Prince and Huguenard, 1988). Indeed, in the few instancesstudied, interneurons have a predominantly excitatory responseto synaptic input (Fig. 7; McCormick et al., 1985). Anatomical evidence suggeststhat singleinhibitory neurons have a stronger impact than individual excitatory ones in determining whether or not a given pyramidal target will fire. GABAergic synapsesare often placed in a position to yield maximum effect, such as on the dendritic shaft, near the cell body, or at the initial segment(Le Vay, 1973; Fair&r and ValVerde, 1980; Fair&n et al., 1983; Beaulieuand Somogyi, 1990). Also, smooth stellate cells may amplify their inhibitory effect by making more synapsesper recipient neuron than do pyramidal cells (Fairen and Valverde, 1980; Somogyi et al., 1983). From our experiments, we concluded that IPSPs evoked by remote stimuli were mediated by more than one synapse.Becausethe arbors of most interneurons are confined to an area about 200-300 pm in diameter (Lorente de No, 1944; Somogyi et al., 1982; Fairen et al., 1983; Naegeleand Katz, 1990) most of the inhibition we saw was probably local. Somebasket cells have axons extending for a millimeter or longer (Martin et al., 1983; Somogyiet al., 1983)and possiblycould have contributed a monosynaptic inhibitory component that wasmaskedby the EPSP. In somepreliminary experiments, however, where the EPSP was pharmacologically blocked, we saw no such component. Last, our resultsshowedthat both fast and slow IPSPs could be produced by sometype or types of cell intrinsic to the superficial layers. That only fast, and not slow, IPSPs were evoked by lateral stimuli may have been a matter of the relative strength of synaptic input onto interneurons. The late IPSP is usually found

1808

Hirsch

and Gilbert

* Synaptic

Physiology

of Horizontal

Connections

to have a higher threshold than the early one (Avoli, 1986; Howe et al., 1987a,b; Dutar and Nicoll, 1988a; cf. Connors et al., 1988). It is possible that the GABA, receptors have a lower affinity for GABA than do the GABA, receptors and therefore interneurons must fire rapidly to release enough transmitter to open the GABA,-linked channels (Newberry and Nicoll, 1985; Dutar and Nicoll, 1988a). Alternatively, it may be the case that different types of neuron selectively produce either fast or slow IPSPs and that horizontal fibers tend to contact the former. Though the evidence is stronger in the hippocampus (Newberry and Nicoll, 1985) than in the neocortex (Connors et al., 1988), it seems that sensitivity to the GABA, antagonist baclofen is greatest in the distal dendrites, whereas responsiveness to GABA, antagonists is widespread. Perhaps some lateral inputs preferentially excite those inhibitory cells, such as basket cells, that synapse on or near the soma (McGuire et al., 1991). Whatever the mechanism, the prevalence of the fast IPSP over the slow one suggests that laterally evoked inhibition can be quickly gated on and elf. Relevance to studies of receptive fields Experiments measuring responses to stimuli placed inside and outside the receptive field provide ideas about the role horizontal connections play in visual processing. One finding of these experiments is that stimulating the surround has dual consequences, in some cases causing excitation, and in most, inhibition. When the surround has an excitatory effect, it cannot drive a cell directly but can facilitate the response to stimuli placed inside the receptive field (Allman et al., 1985; Nelson and Frost, 1985; Orban et al., 1987; Wiesel and Gilbert, 1989; Gilbert and Wiesel, 1990). A mechanism for this could involve the voltage dependence of the laterally evoked EPSP. The EPSP would be small with illumination of the surround only, but would increase in size as stimuli presented within the receptive field recruited additional depolarizing inputs. Frequently, the influence of the surround is suppressive (Bishop et al., 1973; Nelson and Frost, 1978; Orban et al., 1987; Wiesel and Gilbert, 1989; Gilbert and Wiesel, 1990). Our results showed that, as increasing numbers of lateral fibers were activated, progressively stronger inhibition was provided. Thus, even though the horizontal collaterals are extended largely by pyramidal cells, which are presumably excitatory, and in turn make a minority of postsynaptic contacts with inhibitory neurons, the remote inhibitory effects observed in vivo could well be generated by the horizontal connections. Both the facilitatory and the suppressive effects of the surround could stem from the same distant sites since the IPSPs evoked by lateral stimulation are probably of disynaptic, local origin. Because the horizontal connections in area 17 join regions of similar orientation specificity (Ts’o et al., 1986; Gilbert and Wiesel, 1989), the lateral IPSPs and EPSPs might share orientation preference with their targets. Intracellular recordings made in vivo reveal that postsynaptic potentials of both polarities are tuned to the same angle (Ferster, 1986, 1989), and the implication from our studies is that inhibition from remote parts of the visual field would obey a similar rule. The state dependence of the horizontal connections could be responsible for nonlinear interactions between the lateral and interlaminar inputs and may account for the unusual effects of activating the surround on orientation tuning, such as changes in preferred orientation or tuning bandwidth (Gilbert and Wiesel, 1990). One may speculate that the modulatory effects of the

horizontal fibers could be responsible for gating of other synaptic pathways, allowing an expression of some and suppression of others.

References Allman JM, Meizin F, McGuinnes E (1985) Direction and velocity specific surround in three cortical visual areas of the owl monkey. Perception 14:105-126. Artola A, Singer W (1990) The involvement of N-methyl-D-aspartate receptors in induction and maintenance of long-term potentiation in rat visual cortex. Eur J Neurosci2:254-269.

Avoli M (1986) Inhibitorv uotentialsin neuronsof the deeulayersof the in v&o neocortical &es. Brain Res 370: 165-170. - _ Beaulieu C, Somogyi P (1990) Targets and quantitative distribution of GABAergicsynapses in the visualcortexof thecat.Eur J Neurosci 21296-303. Bishop PO, Coombs JS, Henry GH (1973) Receptive fields of simple cells in the cat striatecortex. J Physiol(Land)213:3l-60. Callaway EM, Katz LC (1990) Emergence and refinement of clustered horizontal connections in cat striate cortex. J Neurosci 10: 1134-l 153. Chervin RD, Pierce PA, Connors BW (1988) Periodicity and directionality in the propagation of epileptiform discharges across neocortex. J Neurophysiol 60: 1695-I 7 13. Connors BW, Prince DA (1982) Effects of the local anesthetic QX3 14 on the membrane properties of hippocampal pyramidal neurons. J Pharmacol Exp Ther 220:476-l8 1. Connors BW, Malenka RC, Silva LR (1988) Inhibition of neocortical pyramidal neurons in rat and cat: two types of postsynaptic potentials, and GABA, and GABA, receptor mediated responses. J Physiol (Lond) 406:443-468. Creutzfeldt OD, Garey LJ, Kuroda R, Wolff JR (1977) The distribution of degenerating axons after small lesions in the intact and isolated visual cortex of the cat. Exp Brain Res 27:4 19440. Dutar P, Nicoll RA (1988a) A physiological role for GABA, receptors in the central nervous system. Nature 332: 156-l 58. Dutar P, Nicoll RA (1988b) Pre- and postsynaptic GABA, receptors in the hippocampus have different pharmacological properties. Neuron 1:585-591. FairCn A, Valverde FA (1980) A specialized type of neuron in the visual cortex of cat: a Golgi and electron microscope study of chandelier cells. J Comp Neurol 194:761-779. Fairtn A, DeFelipe J, Regidor J (1983) Nonpyramidal neurons: general account. In: Cerebral cortex, Vol 1, Cellular components of the cerebral cortex (Peters A, Jones EG, eds), pp 201-253. New York: Plenum. Ferster D (1986) Orientation selectivity of synaptic potentials in neurons of cat primary visual cortex. J Neurosci 6: 1284-l 30 1. Ferster D (1989) The synaptic inputs to simple cells of the cat visual cortex. In: Neural mechanisms of perception (Lam DM-K, Gilbert CD, eds), pp 63-85. Houston: Gulf. Gabbott PLA. Martin KAC. Whitteridae D (1987) Connections between pyramidal neurons in layer 5 gf cat visual’cortex (area 17). J Comp Neurol 259:364-38 1. Giffin K, Doyle JP, Nerbonne JM (1988) Comparison of excitable membrane properties in identified neurons from mammalian visual cortex. Sot Neurosci Abstr 14~297. Gilbert CD, Wiesel TN (1979) Morphology and intracortical projections of functionally identified neurons in cat visual cortex. Nature 280:120-125. Gilbert CD, Wiesel TN (1983) Clustered intrinsic connections in cat visual cortex. J Neurosci 3: 1116-l 133. Gilbert CD, Wiesel TN (1985) Intrinsic connectivity and receptive field properties in visual cortex. Vision Res 25:365-374. Gilbert CD, Wiesel TN (1989) Columnar specificity of intrinsic horizontal and corticocortical connections in cat visual cortex. J Neurosci 9:2432-2442. Gilbert CD, Wiesel TN (1990) The influence of contextual stimuli on the orientation selectivity of cells in primary visual cortex of the cat. Vis Res 30: 1689-1701. Hestrin S, Nicoll RA, Perkel DJ, Sah P (1990) Analysis of synaptic action in pyramidal cells using whole-cell recording from rat hippocampal slices. J Physiol (Lond) 422:203-225. Horikawa K, Armstrong WE (1988) A versatile means of intracellular

The Journal

labeling: injection of biocytin and its detection with avidin conjugates. J Neurosci Methods 25: l-l 1. Howe JR, Sutor B, Zieglgtinsberger W (1987a) Baclofen reduces postsynaptic potentials of rat cortical neurones by an action other than its hyperpolarizing action. J Physiol (Land) 384:539-569. Howe JR, Sutor B, Zieglglnsberger W (1987b) Characteristics of longduration inhibitory postsynaptic potentials in rat neocortical neurons in vitro. Cell Mol Biol 7: l-l 8. Hubel DH, Wiesel TN (1963) Shape and arrangement of columns in cat’s striate cortex. J Physiol (Lond) 165:559-568. Hubel DH, Wiesel TN (1967) Cortical and callosal connections concemed with the vertical meridian of the visual fields in the cat. J Neurophysiol 30: 156 l-l 573. Humphrey AL, Sur M, Uhhich DJ, Sherman SM (1985) Projection patterns of individual X- and Y-cell axons from the lateral geniculate nucleus to cortical area 17 in the cat. J Comp Neurol233:.159-189. Innocenti GM (1980) The mimarv visual Dathwav throuah the cornus callosum: morphological and functional aspects-in the cat. Arch Ital Biol 118:124-188. Jones KA, Baughman RW (1988) NMDA- and non-NMDA receptor components of excitatory synaptic potentials recorded from cells in layer V of rat visual cortex. J Neurosci 8:3522-3534. Katz LC (1987) Local circuitry of identified projection neurons in cat visual cortex brain slices. J Neurosci 7: 1223-l 249. Kauer JA, Malenka RC, Nicoll RA (1988) A persistent postsynaptic modification mediates long-term potentiation in the hippocampus. Neuron 1:911-917. Kerr DIB, Ong J, Johnston GAR, Abbenante J, Prager RH (1988) 2-Hydroxy-saclofen: an improved antagonist and at central and peripheral GABA, receptors. Neurosci Lett 92:92-96. Kisvarday ZF, Martin KAC, Freund TF, Magloczky Z, Whitteridge D, Somogyi P (1986) Synaptic targets of HRPlfilled layer III pyramidal cells in the cat striate cortex. Exp Brain Res 64:541--552. LeVay S (1973) Synaptic patterns in the visual cortex of the cat and monkey. Electron microscopy of Golgi preparations. J Comp Neurol 150:53-86. LeVay S, Gilbert CD (1976) Laminar patterns of geniculocortical projection in the cat. Brain Res 113: 1-19. Lorente de No R (1944) Cerebral cortex: architecture, intracortical connections, motor projections. In: Physiology of the nervous system (Fulton JF, ed), pp 29 l-325. Oxford: Oxford UP. Lijwel S, Freeman B, Singer W (1987) Topographic organization of the orientation column system in large flat-mounts of the cat visual cortex: a 2-deoxyglucose study. J Comp Neurol255:401415. Lowe1 S, Bischof H-J, Leutnecker B, Singer W (1988) Topographic relations between ocular dominance and orientation columns in the cat striate cortex. Exp Brain Res 71:33-46. Martin KAC, Whitteridge D (1984) Form, function and intracortical projections of spiny neurones in the striate visual cortex of the cat. J Physiol (Lond) 353:463-504. Martin KAC, Somogyi P, Whitteridge D (1983) Physiological and morphological properties of identified basket cells in the cat’s visual cortex. Exp Brain Res 50: 193-200. Mayer ML, Westbrook GL (1984) Mixed-agonist action of excitatory amino acids on mouse spinal cord neurones under voltage-clamp. J Physiol (Lond) 354:29-53. McCormick DA, Connors BW, Lighthall JW, Prince DA (1985) Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex. J Neurophysiol54:782-806. McGuire B, Gilbert CD, Wiesel TN, Rivlin PK (1991) Target morphology and specificity of horizontal connections of pyramidal neurons in the primate visual cortex. J Comp Neurol, in press. Morrison JH. Foote SL. Molliver ME. Bloom FE, Lidov HGW (1982) Noradrenergic and serotonergic fibers innervate complementary lay: ers in monkey primary visual cortex: an immunohistological study. Proc Nat1 Acad Sci USA 79:2401-2405. Naegele JR, Katz LC (1990) Cell surface molecules containing N-acetylgalactosamine are associated with basket cells and neurogliaform cells in cat visual cortex. J Neurosci 10:540-557. Nelson JI, Frost BJ (1978) Orientation-selective inhibition from beyond the classic visual receptive field. Brain Res 139:359-365. Nelson JI, Frost BJ (1985) Intracortical facilation among co-oriented

of Neuroscience,

June

1991,

1 I(6)

1809

co-axially aligned simple cells in cat striate cortex. Exp Brain Res 6 1: 54-61. Newberry NR, Nicoll RA (1985) Comparison of the action of baclofen with gamma-aminobutyric acid on rat hippocampal pyramidal cells in vitro. J Physiol (Lond) 360:161-185. Orban GAB, Gulyas B, Vogels R (1987) Influence ofa moving textured background on direction selectivity of cat striate neurons. J Neurophysiol 57:1792-1812. Prince DA, Huguenard JR (1988) Functional properties of neocortical neurons. In: Neurobioloav of neocortex (Rackic P. Singer 1 W.I eds). II--nn 153-l 76. New York: Wyley. Rockland KS, Lund JS (1983) Intrinsic laminar lattice connections in primate visual cortex. J Comp Nemo1 2 16:303-3 18. Rockland KS, Virga A (1989) Terminal arbors of individual “feedback” axons proiecting from area V2 to Vl in the Macaque monkey: a study using immunohistochemistry of anterogradely‘transported Phaseolus vulgaris-leucoagglutinin. J Comp Neurol285:54-72. Schwartzkroin PA, Mathers LH (1978) Physiological and morphological identification of a nonpyramidal hippocampal cell type. Brain Res 157: l-10. Shatz CJ (1977) Anatomy of interhemispheric connections in the visual system of Boston Siamese and ordinary cats. J Comp Neurol 173:497-518. Shirokawa T, Nishigori A, Kimura F, Tsumoto T (1989) Actions of excitatory amino acid antagonists on synaptic potentials of layer II/ III neurons of the cat’s visual cortex. Exp Brain Res 78:489-500. Sillito AM (1884) Functional considerations of the operations of GABAergic inhibitory processes in the visual cortex. In: Cerebral cortex, Vol 2, Functional properties of cortical cells (Jones EG, Peters A, eds), pp 91-117. New York: Plenum. Somogyi P, Freund TF, Cowey A (1982) The axo-axonic interneuron in the cerebral cortex of the rat, cat and monkey. Neuroscience 7: 2577-2607. Somogyi P, Kisvarday ZF, Martin KAC, Whitteridge D (1983) Synaptic connections of morphologically identified and physiologically characterized large basket cells in the striate cortex of cat. Neuroscience 10:261-294. Staftstrom CE, Schwindt PC, Chubb MC, Crill WE (1985) Properties of persistent sodium conductance and calcium conductance of layer V neurons from cat sensorimotor cortex in vitro. J Neurouhvsiol53: - _ 1153-1170. Sutor B, Hablitz JJ (1989a) EPSPs in rat neocortical neurons in vitro. I. Electrophvsioloaical evidence for two distinct EPSPs. J Neuronhvs- _ iol 613621634. Sutor B, Hablitz JJ (1989b) EPSPs in rat neocortical neuron in vitro. II. Involvement of N-methyl-D-aspartate receptors in the generation of EPSPs. J Neurophysiol 61:607-620. Thomson AM, Grirdlestone D, West DC (1988) Voltage-dependent currents prolong single-axon postsynaptic currents in layer III pyramidal neurons in rat neocortical slices. J Neurophysiol60: 1896-1906. Tigges J, Tigges M, Anschel S, Cross NA, Letbetter WD, McBride RL (198 1) Area1 and laminar distribution of neurons interconnecting the central visual cortical areas 17, 18, 19, and MT in the squirrel monkey LSaimiri). J Como Neurol 202:539-560. Tovama K (1988) Functional connections of the visual cortex. In: Neurobiology of neocortex (Rackic P, Singer W, eds), pp 203-217. New York: Wiley. Toyama K, Kimura M, Tanaka K ( 198 1a) Cross-correlation analysis of intemeuronal connectivity on cat visual cortex. J Neurophysio146: 191-201. Toyama K, Kimura M, Tanaka K (198 1b) Organization of cat visual cortex as investigated by cross-correlation technique. J Neurophysiol 46:202-2 12. Ts’o D, Gilbert CD, Weisel TN (1986) Relationships between horizontal connections and functional architecture in cat striate cortex as revealed by cross-correlation analysis. J Neurosci 6: 1160-l 170. Watkins JC, Evans RH (198 1) Excitatory amino acid transmitters. Annu Rev Pharmacol Toxic01 2 1: 165-204. Wiesel TN, Gilbert CD (1989) The Helmerich lecture: neural mechanisms of visual perception. In: Neural mechanisms of perception (Lam DM-K, Gilbert CD, eds), pp 7-33. Houston: Gulf.